foreign ation maybe and also the different Technologies of AI and Quantum Computing and how they shape the digital landscape then Professor Dr zabina yeska your state ladies and gentlemen very happy to be here with you today um and I would like to talk about two of my favorite topics which is artificial intelligence I'm doing that since about 20 years or so and the topic of quantum Computing where I have been digging in in about 2015. let me give you just a very brief introduction why is it so important to consider both sides of these medals so to say we are in a fourth Industrial Revolution if you shorten the first Industrial Revolution to just one word what would it be dump machines steam engine if you shorten the force Industrial Revolution just to one phrase what would that be it's AI it's this first Industrial Revolution is a breakthrough of artificial intelligence it's a revolution where technological things get intelligent and what we face today is more or less the algorithmic driven face and we
have seen this enhanced talk it's about reinforcement learning deep learning Q learning are learning supervised learning and whatever learning and we are in the algorithm driven face but we will see that there are at least two other Technologies which will Skyrocket Artificial Intelligence on the next levels one is connectivity 5g6g and this is because in these protocols the latency time is so short that consider also namaskar where do you put the computer right now with 4D in the car where would you put the computer with 5g6g and some kind of a cloud because it time the latency shot is so short that's much better to use a much more advanced machine within the in the car you always have limitations to space to energy consumption heat and so on so in particular in AI form autonomic systems the second phase which is going to break through now 5G is around six years 2028 about this will happen now and then there is HPC country Quantum Computing which again brings AI to the next level and
this is my my talk for today and I have kind of four chapters I first would like to get off some wrong assumptions around Quantum computing and then I would like to from a helicopter perspective show you I mean we already have a lot of computational power right so why do I need that I mean what's what's really changing and then from there I would like to show you examples and finally I would like to couple the topic of quantum Computing with sustainability and the climate change so this will be my outro I don't have to give a summary at the end wrong assumptions number one oh I have so many other things to do I can't care about Quantum Computing because it's very long way to go so Quantum measures very far away you always hear that if you do uh Consulting in this field as I do well look at these numbers by 2025 you will see the Deep temperature Quantum Computing stack breaking through on Industrial scale in 2028 it's the room
temperature systems you don't have to cool so if we consider I mean taking particular German Industries how long it has been taking us to understand accept and integrate AI blockchain a lot of other Technologies now we have two years to go okay start today second wrong assumption I'm just talking about the left example today because I have been a board member for Roger Barn a couple of years I don't need Quantum Computing because you only need it if you do medical research protein folding or climate investigations or something like that oh this is wrong again consider only the tricks the stations and the trains ignore maintenance ignore people ignore everything else just trains tracks and stations write a simulation and come up with a schedule computational time would be about two or three weeks which means that this is nice to have and you can use it to understand to do analytics understand about problems and then come up with a better schedule for next year but you cannot use it to steer your system
in real time but this is what you need we have about like three four disturbances per minute train is leaving a station a minute too late because of too many people a group of animals is running on the tracks we have to stop and so on and so forth so it's a very basic question to have more high performance Computing than we have today it's not only for special applications third I cannot do Quantum Computing because there are no quantum computers today and while this is wrong once again so first of all you can prepare your organization hire the right people integrate yourself into ecosystems find out about the best use cases perhaps you have some governance problems for example bringing all the data together because you want them to use them integratedly so on the organizational side there's a lot of stuff to do but you also can already use certain quantum computers in particular the annealers it's kind of a in-between technology or you can go for Quantum inspired algorithms which basically means
that you program in a way you would do it for a quantum computer since it's not around you use traditional computers but later on you just flip the handle and move number four uh is oh I think I I skipped this from my for my slides but news number four has been it's too expensive well now you know whenever a new technology breaks through it's kind of expensive but the first quantum computers will be deep temperature Quantum Computing it will be in clouds you will not buy it you will not own it you will have a license for it and by that it's getting affordable and then later on some more of these computers are showing up you will not face a problem any longer so now what is happening when quantum's Computing Quantum Computing breaks through this is a bit more philosophical slide perhaps for the only one so um they have been certain incidents in the development of Hardware technology for computer science which kind of we had a revolutionary evolutionary development of
Hardware over the last 70 or so years but at certain points in time there was something more and this was when we freed the computers for the from the traditional computer centers and came up with the PC um from the physics inside it was not a big difference it was just smaller but before the computers were just used by a very small group of people some researchers some other people highly educated and then by then by having a PC for everybody we opened up for all people for diversity for all branches for our applications and we got a revolution within the digitalization without changing very much about the hardware it was just that we changed the access so it was kind of a freedom to give the computer give people freedom to use it or give the computers the freedom to be used by anybody this happened later on with the mobile internet once again now it's not about people but so far computers have been mainly mainly stationary now with the mobile internet you
could have the computers anywhere everywhere so we built in a lot of computers into tracks to understand where our trains are for example so again this was a degree of freedom and now it's more geographically and what's happening now is there's so much limitations today things you would like to compute but cannot with Quantum Computing of course you cannot do everything and in particular not in the first years but in general you will see that it's really opening up what you can do and this is perhaps my one but next slide for one more point before I come to that moose law is coming to its end it might be 2025 it might be 2030 because perhaps we come up with some you know tricks but in general because of the limitation of the miniaturization it's coming to its end we are so much used to Moore's Law if I buy a mobile phone now I will buy one after two years I expect it to be different to be better not only in terms
of computational power but also in terms of computational power so we are all so much used to the smallest law computers are just getting better somehow that if this is no longer the case it feels like being frozen in in the innovation of computer science so it's really good that Quantum Computing is showing up now by 2025 and it will not only kind of be a success of Muslim will be more like an explosion in at least four fields we will be capable of one of our number one of doing totally new things things where I could have written the simulation for it but I would know that to execute it it would take 20 000 years by then I would be dead and the computer would be broken so in general I can write simulation but in fact I cannot execute it these are very very big models not simplified models any longer for the next years I believe that number two number three is most important my Gogan back to my Deutsche Bank
example we have the schedule planner now but you cannot use it in real time by now with Quantum we can and that means that we can use these kind of schedules to really steer the system in real time and trying to whenever something happens to bring the system back on schedule as fast as possible today it's done by humans and of course I mean if you have a problem in Hamburg these humans are very good but they can only optimize for Hamburg so whatever they decide you have no clue what's happening in Paso number three Optimal Solutions very often we give a computer for for optimization problem a certain amount of time and then within this time we take the best solution we came up with if this is the worst but acceptable solution this is your Optimum sometimes you get a solution here and everything else here is waste but we don't know about it so I believe that we will come up with much much better resource usage by using quantum computers and
then finally product design we will run we will have simulations based on physical correct digital twins we will take the simulation clone it 1000 times speed up the speed and then run simulations in such a way you can never do it in a real lab and by that we will come up with a acceleration of Innovations in particular everywhere but in particular if it comes to the atomic scale stuff chemistry drugs design Medicals and so on two more slides for this chapter I have been saying that there is a cold world of quantum Computing and there is the other one the room temperature and even within I'm going back come here even Within These post World there are different approaches how to design this Hardware as you can see from the turquoise Bubbles and then there are different companies trying to realize that and they're doing it in a different way and this is just kind of a screenshot of I did the couple of weeks ago so what I want to say is here
and also here it's a zoo big companies small companies traditional companies new companies different Technologies warm cold in between um we so far we do not know which kind of Technology will be successful there are certain members of the community which say well it will be exactly this one super fluidity for example others are more open my personal expectation is that in the future we will have much more different Technologies in the computer center because the different Hardwares might be more suited for one or the other problem so I expected to a certain degree we will have not perhaps this kind of zoo a limited zoo in the computer centers so examples whenever I come into a company and we agree to agree that Quantum Computing is important right now then the question starts okay Sabina we understand now Quantum is important you can use it kind of for a lot of questions where should I start good question and what I always use is this kind of slide saying yes perhaps in the future
you can use it for kind of everything but now start with where are you doing or want to do optimization where do you do simulation where do you do or want to do AI of course very often these pillars are either tackled because I mean you have a simulation with a lot of optimization in between and then part of the optimization might be Ai and the other one might be traditional mathematical stuff however think about optimization AI simulation and now goes through the rows General processes do you have for example an optimization problem in Human Resources like bringing all the locomotive drivers on their locomotives in the morning something like that General processes are HR procurement finances mainly this is a very nice point to start because here you can start working together with other companies because very often I mean let's say you have Volkswagen you can work together with BSF you're not in a competition because you're doing totally different stuff but the solutions to pick the right people or to bring the
right people to the right moment and do the right plans is a very similar one specific processes means you have a product how do you produce it can you use quantum Computing to optimize your production this is for example true for aim automobile companies during the robots I will see this in a second and then also for the product itself let's say you are a bank you are coming up with some Robo advisor uh could this robot advisor perhaps be better if it's using Quantum computing if you go through this three course three Matrix you will perhaps not find all the interesting use cases but I guarantee for 80 to 90 percent I'll show you some and these examples um I am not picking them because I found them the best in the world but because I already realized because I want to show you that this is not a science fiction talk so this is IBM and x and mobile and they had the problem that for the liquid gas you have a lot
of ships you have these gas Fields you have a lot of harbors and then you have the silos for the intermediate storage and so on that they were not happy with their routing concept when a customer comes and says I want more or less of the liquid gas then usually the ship goes to a different gas field the slot in the harbor is going and so on and so forth there are so many parameters already in this question that they have been using annealers to optimize a solution as any less in contrast to Traditional High Performance Computing and came up with very nice results it's about a year olds as example year three mathematically spoken this is quite the same because these are all variations of a traveling salesman problem here you have to repair something stationary in this case a plane and then you have to get the right spare parts at the right place where the right people with the right specialist in a certain order because certain things cannot be done before
other things have been done and so on and so forth and of course you don't want to have too much storage and you don't want to have the Specialists there but you cannot use them because you need them in five days so this kind of simulations can be very very nicely done this was with a d-wave vanilla next example is actually the finance branch has been starting pretty early with Quantum Computing or at least with investigations they did already the same with high frequency trading you might remember a couple of years ago about seven or ten or so because I mean if you trade shares then you have a lot of data points considering the business industry but also other aspects like political data points weather data points whatever and the better you can integrate them into a big big simulation see more precise your result might be concerning where the share is going to go however if you take too many of these points and your computational title is too long and you cannot
react fast enough so in finance is a lot of stuff is going on using all data points in the world to make a very good prediction for which shares going where um but do it very very quick and this example again for detection using Quantum AI is mathematically spoken the same because you cannot avoid to be hacked the only thing you can try to do is if you get hacked to figure out about it as fast as possible like the immune system of a healthy body and then isolates the problem as fast as possible so again it's a question of analyzing the whole structure of your I.T system figuring out oops there's a pattern I don't like and then close the walls and it's a question of time and those are two examples I have been borrowing from my colleague Floyd who's now with Tara Quantum which is a company on Quantum Computing but at this time he was with Volkswagen so I'm using this logo look at this four pictures for a second and
for a human it's very easy to understand that it's a town X Y coordinates are the same it's just that the weather conditions are changing and they enter so one is night one is day right for artificial intelligences this kind of image recognition is very difficult it can be done but it takes a time and because it's so slow it this feature cannot be part of autonomous driving today but we are using this feature all the time oops I've been here before is something we use to drive a car so we cannot have this feature oops in a car today but we can using quantum and this is my my last example again coming back to AI very straight reinforcement learning why do we don't use more reinforcement learning you know this technology you don't need data you just let the system try around hoping for finding its goal um it is a very nice approach very very powerful very flexible and you don't need data which is great because of you all the privacy
issues the problem with reinforcement sometimes is that if you have a very high parameter space then the system starts to play usual with it's a it's paint brushes instead of painting the hair because it's allowed to do everything so reinforcement sometimes has a barrier of being slow at the teaching process if you can speed it up and this is what Florian showed that if you can if you use the speed of quantum systems you can speed reinforcement algorithms up in such a way that they can be used in production technology avoiding all this teach and stuff and so on um last last few minutes there is a topic nobody talks about everybody's just talking about ah Quantum Computing the new HPC and so on but Quantum Computing will also contribute it will not solve all the way it will contribute to sustainability and this is for three reasons first of all because we can be much faster in innovating new technologies conversion example in second second because optimization of resources usually means optimization not only
in terms of costs or human resources it also means in terms of something with clear climate issue climate impact and certainly we have to consider the energy consumption of quantum computers compared to traditional ones and uh for the first example Issa European special agency is working on power plants based in the orbit if you could do that then uh of course if you put elements with the right orbit that would be signed at any time 24 7 and that means that you could have green energy not only by day but also by night it was about 25 of these Fabrics you could come up with the energy consumption for Europe 25 might sound big but if you know that we have much more than 100 nuclear power plants then 25's close to nothing so this kind of Fabrics or this kind of big engineering tasks usually take for decades and now comes what I have been talking before if you can enhance the Innovation speed coming up with how do I do that and how
do I test it and how do I whatever in our 30 years but three years could really make a difference my second example for optimization always I mean optimization everybody thinks about okay I have a truck I have a fleet of trucks and if I shorten traveling salesman ever shortens away then I shorten the diesel consumption okay it's too much boring this example is much nicer when you take a plane you know that the burning of kerosene is not so fantastic for the climate but actually it's only one third of the problem about 50 of the problem come from these condensation trails these are artificial clouds produced by the airplane and they usually are more warm than the environment and therefore they heat the environment and this is exactly what you don't want in the atmosphere so Ken shows there's a very nice study from dll two or four years ago showing that if you change the pairs of the plane in all three dimensions x y and z slightly you can move it in
such a way through the atmosphere that the um the difference in the temperatures is not as much as it is usually and by that you can reduce the condensation trails sometimes you can even reverse the effect so you can use a plane to cool the atmosphere which is nice but that's only a small number of cases it happens if you do that you have to recompute the pairs of the plane by every second seconds 30 seconds and of every plane and then the independence between the different planes and we did a very very nice challenge with Deloitte on the AI pack and a lot of startups showed up and showed how you can use quantum Computing to solve this riddle we did estimations you cannot do it with the traditional software Hardware around but you can with quantum and then my very last slide to also stay and time a lot of people believe that Quantum Computing will take up a lot of energy this is firstly because high performance Computing always takes up a
lot of energy and second because I have heard oops I have to cool the system down to minus 273 degrees this is very cold and this needs of course a lot of of energy but the truth is the other way around the systems themselves due to the physics and we're going to have a physics class today but due to the physics of the system the computers themselves needs only a very very small amount of energy now for of course the code words the Deep temperature systems you have to cool them but we have them in specialized environments a lot of different computers in the same environment sharing the same laboratory choice and so on so the Assumption of the specialist is that it will be about 10 percent compared to what we need for today and if it comes to room temperature there's no cooling the Assumption we cannot measure it I mean this is based on assumptions physics talking to the hardware guys but we assume that we are going down for two magnitudes
two to one two to five percent as my last sentence if this happens the right one then this changes the game totally because right now even me and my talk I have been presenting Quantum Computing as a solution for sophisticated problems but if this is true then I would like to Swatch my iPhone to a q phone not because I really need this kind of computational power in a mobile phone but because of sustainability reasons because the energy consumption is so much lower let's work on that thank you hi Sabina so I have one question yes um there exists some quantum computer but not enough right and I think people for people it's not tangible enough now and I know you always say yes it's not there now but you need to start now and to have this in like two two years for example um so um the thing is that we we need to make these topics a bit more tangible to grab right and uh for example with AI there was this
happening with ChatGPT um and now for every person who's not Tech related it's clear oh AI is a thing what could be the chat CPT happening for Quantum computing um actually it might perhaps be the same chatty BT you know what I did on the weekend I told jgbt please program traveling salesman problem five arbitrary cities in C okay it did I told him okay what do you know about kiss kit a kiss kit is one of the languages for Quantum Computing and taught me I can do everything okay program it and kiss kit and I got the kisket code so perhaps um there are two barriers I mean the first barrier is where's the hardware and the second barrier is you cannot program them in traditional languages I think the second one we are about to solve and the first one if you run your data center on AWS clouds for example they already have been starting to offer accessibility to quantum computers around in the world of course the majority of them are
kind of babies to toddlers most of the baby side perhaps so it's more for proof of concept but you can start to educate your people let them try around figure out about who of your it guys is really interested in this kind of stuff so it can be solved you also I always say like there are a lot of routes to Rome always um and the quantum computer finds the faster and more most efficient one right um couldn't we also use a Quantum Computing to bring AI to the highest level uh definitely I mean we see that in my last last part one example that this reinforcement topic is really important for Europe because um a lot of newspapers start with something like industry says I can't do AI because I don't have the data if you do the reinforcement you don't need the data and then people say well yeah but I don't know about reinforcement and by the way it's too slow so in this way it can be a big big help
this is just one example I believe it will be a big help in other areas as well I've seen one hand over there do we have a microphone Maybe um um yeah one second of course we have one and in the online audience please also dare to question something so who's there uh hello Hi how are you my name is Joanna Bryson uh thank you for your talk I have uh two questions I I admit they'll be challenging because the parts I liked are disliked I don't have questions about um one wasn't really yours but you mentioned it I'm terrified of the idea of real-time control of automobiles being done over 5G because what if somebody disrupts it if there's a Cyber attack so and I think this sort of relates to your talk because that's I suppose one of the things you hope Quantum will be doing is providing some of the servers and then secondly uh the claims you made on the final slides about what's going on with power I admit I
haven't kept up the last few years but the last I knew people were saying that the power usages of quantum are so extreme that only like uh again security is good to be able to afford it so what are the current empirical numbers for for uh for the uh uh power usage yeah and what when do you see these kinds of returns that you're talking about happening okay so that was the second question so I don't have a study right now but working on it actually and this are kind of free results from a lot of communication I had with people from different of the companies so this is the best um estimation I can give you right now um for the first question uh the real-time question of autonomous cars so first of all um if you take for example full breakage system then it's very very important that this reaction is as fast as possible obviously so um all Concepts like having a human in the loop or something like that would not
work in such a situation so basically we come down to the fact do we want to have autonomous cars or not if we want them then they have to be as secure as possible and the architecture for that will be to have kind of a little brain within the car which is capable of oops there is a whatever barrier I have to break and then you have to have the big brain in the cloud and usually you will be you will compute in the cloud but when the connection is not there or when there is a kind of you know this system says a but this system says C then okay start again something like that so this will be the solution for that I believe another short question Maybe yeah was that a hand here yeah in the second row was that a hand no okay no that's fine okay so you're still here Savina um we would like to have a short bio break now but please be there in time quarter two
and then we will talk about weather Germany can be a technology leader and how so this is interesting and important for everyone um and thank you a big Applause for Sabina thank you